Monitoring Disk IO on HA Cluster via SNMP

0ddba11 · 09-12-2014, 06:46 AM

Hi all!

We use Cacti to monitor various server metrics and I have recently started looking into using it for disk IO monitoring.

The method found uses the UCD-DISKIO-MIB SNMP MIB which unfortunately seems to use the dynamically created device mapper nodes (/dev/dm-##).

This of course is not only subject to change, but isn't the same on both nodes of the cluster when the service migrates to it.

In fact, according to Red Hat's documentation...

Quote:

NOTE: /dev/dm-N nodes should never be used in scripts to access the device since the N number is assigned dynamically and changes with the sequence of how devices are activated. Therefore, true names in the /dev/mapper directory should be used. This layout is to support udev requirements of how nodes/symlinks should be created.

So my questions are:

Is there a way to configure snmpd to report the /dev/mapper entries which aren't subject to change instead of the device mapper nodes?
Or instead, is it possibly to get UDEV to use the same persistent device mapper node for the volume in question on all nodes of the cluster?

TB0ne · 09-24-2014, 09:55 AM

Quote:

Originally Posted by 0ddba11

Hi all!
We use Cacti to monitor various server metrics and I have recently started looking into using it for disk IO monitoring. The method found uses the UCD-DISKIO-MIB SNMP MIB which unfortunately seems to use the dynamically created device mapper nodes (/dev/dm-##). This of course is not only subject to change, but isn't the same on both nodes of the cluster when the service migrates to it.

So my questions are:

Is there a way to configure snmpd to report the /dev/mapper entries which aren't subject to change instead of the device mapper nodes?
Or instead, is it possibly to get UDEV to use the same persistent device mapper node for the volume in question on all nodes of the cluster?

Well, I think you've got a chance of doing it via modification of the UDEV rules, as you said...but I'd hesitate to do so. A custom udev ruleset may get vlown away by an update, and it's just one more bit of custom-code you have to keep track of when you do migrations.

There is a script that was posted to the Cacti forums a while back, that may give you a start:
http://forums.cacti.net/viewtopic.php?p=124793

I *THINK* you'd have to modify the 13-dm-disk.rules file to get what you're after.